”AI Answer Evaluator”
Mini Project Progress Seminar
By
Sathvik Shetty
Vignesh Vane
Prajwal Lolekar
Name of the Guide
Department of Electronics and Computer Science
Pillai HOC College of Engineering and Technology, Rasayani
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 1 / 17
Outline
1 Introduction
2 Motivation
3 Literature Survey
4 Problem Statement
5 Objectives
6 Design Methodology
7 Implementation
8 Results
9 Conclusion
10 References
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 2 / 17
Introduction
Automated Answer Evaluation: This project focuses on developing
an AI-based system to assess subjective answers with high accuracy[1].
Reducing Manual Effort: By detecting the simple digital text, the
system minimizes human intervention in grading, making evaluations
more efficient.
Ensuring Fairness & Consistency: AI-based evaluation eliminates
bias and maintains uniformity in scoring across different responses [2].
Enhancing Learning Systems: The project aims to improve digital
education by providing instant and reliable feedback to students.
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 3 / 17
Motivation
Automating Answer Evaluation – Eliminates manual grading
delays, bias, and inconsistency, making assessment faster and more
objective.
AI-Powered Accuracy – Uses YOLO for answer detection and OCR
for text extraction, ensuring fair and precise evaluation.
Scalable Efficient – Can evaluate answer sheets instantly, reducing
workload for educators and supporting online learning platforms.
Education – Enhances remote learning, integrates with EdTech
platforms, and enables personalized feedback for students.
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 4 / 17
Literature Survey
Traditional vs. AI-Based Evaluation: Research shows that manual
grading is time-consuming and subjective, whereas AI-driven grading
enhances efficiency.[4]
Role of LLM in Answer Assessment: Studies indicate that Large
language Model (LLM) is effective in understanding and scoring
textual responses.[3]
Machine Learning for Pattern Recognition: Prior work
demonstrates how ML models can recognize answer structures and
patterns for better grading accuracy.
Challenges in Automated Grading: Research highlights challenges
like understanding contextual meaning and handling diverse writing
styles.[5]
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 5 / 17
Literature Survey
Figure: Flowchart of Answer Evaluator
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 6 / 17
Problem Statement
The manual evaluation of subjective answers is time-consuming,
inconsistent, and prone to human biases. This project aims to develop
an AI-driven answer evaluation system that leverages Large Language
Model (LLM) and Machine Learning (ML) techniques to assess
responses accurately, ensuring fairness, efficiency, and scalability in
grading.
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 7 / 17
Objectives
To automate the evaluation of printed answer sheets using object
detection and OCR techniques.
To accurately detect and extract question-answer regions from
scanned documents.
To map extracted answers to their respective questions for reliable
assessment.
To integrate the evaluation system into a backend API for seamless
and scalable deployment.
To enhance grading speed and consistency while reducing manual
errors in descriptive answer evaluation.
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 8 / 17
Design Methodology
Software Requirements:
React Native – for mobile frontend development.
Flask – backend server to handle all processing logic.
Python – for running YOLO, OCR, and LLM models.
Firebase – to store uploaded sheets and results.
Model Selection and Processing:
YOLO: Detects answer regions in the uploaded image.
Custom OCR: Extracts handwritten text from detected regions.
Custom LLM: Evaluates answers based on the provided answer key.
PDF Generator: Creates final result PDF with scores and feedback.
UI/UX Design Principles:
User-friendly interface for uploading images and selecting answer keys.
Visual feedback with bounding boxes for detected regions.
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 9 / 17
Implementation
Figure: Block diagram of Answer Evaluator
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 10 / 17
Result I
Extracted Text, Detected Question-Answer Pairs, and Score Allocation
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 11 / 17
Result II
Detected Q&A Regions with Confidence Scores for Accurate Extraction
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 12 / 17
Result III
Marks Assigned to Each Answer Based on Comparison with Reference
Answers
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 13 / 17
Result IV
PDF Format after download
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 14 / 17
Conclusion
The Answer Evaluator Project presents a major advancement in
automating answer sheet evaluation using object detection, OCR, and
LLM. It accurately identifies and pairs printed answers with questions,
enabling consistent, fast, and objective grading. While currently
limited to printed text, the system is scalable and shows potential for
future enhancements like handwritten text support, advanced NLP
models, and cloud-based deployment, making it a valuable tool for
modern educational assessment.
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 15 / 17
References I
1 [1] Raghava Prasad C., Kishore P.V.V., Morphological differential
gradient active contours for rolling stock segmentation in train
bogies,2016, ARPN Journal of Engineering and Applied Sciences, Vol:
11Issue: 5, pp: 2799 - 2804, ISSN 18196608
2 [2] Hari Priya D., Sastry A.S.C.S., Rao K.S., FPGA based design and
implementation for detecting Cardiac arrhythmias ,2016, ARPN
Journal of Engineering and Applied Sciences, Vol: 11, Issue: 5,
pp:3513 - 3518, ISSN 18196608
3 [3] Ur Rahman M.Z., Mirza S.S., Process techniques for human
thoracic electrical bio-impedance signal in remote healthcare systems
,2016, Healthcare Technology Letters, Vol: 3, Issue: 2, pp: 124 - 128,
ISSN 20533713
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 16 / 17
References II
4 [4]Sinha, R., et al., ”Automated Evaluation System Using Object
Detection and OCR,” Journal of Intelligent Systems, 2021.
5 [5]Kumar, A. Shah, P., ”A Scalable Backend API for Image-based
Answer Evaluation,” International Conference on Artificial Intelligence
Applications, 2022.
Prajwal Sathvik Vignesh ”AI Answer Evaluator” July 18, 2025 17 / 17
